julep

Name: julep
Author: julep-ai

byjulep-ai· 1 MCP server

Stars

6.6k

Forks

971

How to install julep?

julep-ai/julep

$git clone https://github.com/julep-ai/julep

Installs into the current project.

›Prefer a prompt? Paste this to your agent

Install & use

Install julep by running `git clone https://github.com/julep-ai/julep`, then use it for the current task and follow its documentation at https://github.com/julep-ai/julep.

Files · 1

View on GitHub

README.md

1	# Julep
2
3	Julep — durable, composable AI agents. Flows that crash and resume, retry safely, and explain every step.
4
5	Julep builds agents as composable, durable dataflows instead of ad-hoc loops: flows can crash and resume, retry safely, explain every step through a derived projection, and deny any tool the model was not explicitly allowed to call. The primary authoring surface is define-by-construction `@flow`: ordinary Python names graph steps while registered tools, pures, reasoners, branches, fan-out, retries, and timeouts compile to the same frozen wire-format IR. The pure core stays dependency-free, while the Temporal layer is optional.
6
7	---
8
9	## Install
10
11	```bash
12	pip install --pre julep
13	```
14
15	Julep 3 currently ships as a release candidate, so the `--pre` flag is required; it drops once `3.0.0` is final.
16
17	---
18
19	## Quickstart
20
21	10 minutes, no API key. Install the base package and run this as a normal Python script:
22
23	```python
24	from typing import TypedDict
25
26	from julep import Reasoner, deploy, flow, pure, think, tool
27
28
29	class SupportReply(TypedDict):
30	reply: str
31
32
33	@tool(effect="read", idempotent=True)
34	def lookup_ticket(ticket: str) -> dict[str, str]:
35	return {
36	"ticket": ticket,
37	"queue": "billing",
38	"summary": "Use the duplicate-charge runbook.",
39	}
40
41
42	@pure("ticket_prompt")
43	def ticket_prompt(hit: dict[str, str]) -> dict[str, str]:
44	return {"queue": hit["queue"], "context": hit["summary"]}
45
46
47	support_reply = Reasoner(
48	name="support_reply",
49	model="anthropic:claude-haiku-4-5-20251001",
50	system="Draft one concise support reply as JSON.",
51	reply=SupportReply,
52	)
53
54
55	@flow
56	def triage(ticket: str) -> dict[str, str]:
57	hit = lookup_ticket(ticket, retries=2, timeout_s=5)
58	prompt = ticket_prompt(hit)
59	answer = think(support_reply, prompt, timeout_s=10)
60	return hit \| answer
61
62
63	def fake_support_reply(value: dict[str, str]) -> SupportReply:
64	return {"reply": f"{value['queue']}: {value['context']}"}
65
66
67	deployment = deploy(triage, tools=[lookup_ticket], reasoners=[support_reply])
68	result = deployment.dry_run(
69	"Customer was charged twice.",
70	reasoners={"support_reply": fake_support_reply},
71	)
72
73	print(result.value)
74	```
75
76	`@flow` runs once at definition time with data handles. Registered tools, registered pures, `think(...)`, `cond(...)`, `switch(...)`, `each(...)`, and `reschedule(...)` append graph steps instead of doing runtime work; `\|` merges records and `h["key"]` plucks fields. `deploy(..., tools=..., reasoners=...)` freezes the tool and reasoner surface, and `dry_run(...)` executes locally with in-memory tools and deterministic fake reasoners. See the larger `@flow` examples in `examples/episode_summary_flow.py` and `examples/cluster_labeling_flow.py`.
77
78	---
79
80	## The CLI
81
82	`julep` is the developer CLI for a whole module of agents — "dbt for agents, terminal-native." Point it at a directory; it discovers every `@flow`/`Agent(...)`, treats each as a node in a cross-agent graph, and gives you one selection grammar across every verb:
83
84	```bash
85	julep ls # list agents (name · kind · tags)
86	julep show triage # one agent's kind, source location, tags, calls
87	julep graph # the cross-agent DAG as Graphviz DOT
88	julep run triage --input '"TICKET-42"' # execute locally, stream the trace tree
89	julep lint +triage # validate an agent and everything it depends on
90	julep test triage # run pytest for the selected agents
91	julep trace <run-id> # render a cached run's trace tree + Langfuse link
92	julep doctor # preflight: discovery, git, Langfuse, Temporal
93	julep deploy triage --env staging # freeze → publish → record in the deploy ledger
94	```
95
96	Selectors compose: `tag:support`, `state:modified` (Slim-CI), `+agent`/`agent+`/`@agent` graph traversal, `a,b` intersection, `--exclude`. Full reference: **[docs-site/content/doc

6 picks

Type

julep

byjulep-ai· 1 MCP server

Stars

6.6k

Forks

971

How to install julep?

julep-ai/julep

$git clone https://github.com/julep-ai/julep

Installs into the current project.

›Prefer a prompt? Paste this to your agent

Install & use

Install julep by running `git clone https://github.com/julep-ai/julep`, then use it for the current task and follow its documentation at https://github.com/julep-ai/julep.

Files · 1

View on GitHub

README.md

1	# Julep
2
3	Julep — durable, composable AI agents. Flows that crash and resume, retry safely, and explain every step.
4
5	Julep builds agents as composable, durable dataflows instead of ad-hoc loops: flows can crash and resume, retry safely, explain every step through a derived projection, and deny any tool the model was not explicitly allowed to call. The primary authoring surface is define-by-construction `@flow`: ordinary Python names graph steps while registered tools, pures, reasoners, branches, fan-out, retries, and timeouts compile to the same frozen wire-format IR. The pure core stays dependency-free, while the Temporal layer is optional.
6
7	---
8
9	## Install
10
11	```bash
12	pip install --pre julep
13	```
14
15	Julep 3 currently ships as a release candidate, so the `--pre` flag is required; it drops once `3.0.0` is final.
16
17	---
18
19	## Quickstart
20
21	10 minutes, no API key. Install the base package and run this as a normal Python script:
22
23	```python
24	from typing import TypedDict
25
26	from julep import Reasoner, deploy, flow, pure, think, tool
27
28
29	class SupportReply(TypedDict):
30	reply: str
31
32
33	@tool(effect="read", idempotent=True)
34	def lookup_ticket(ticket: str) -> dict[str, str]:
35	return {
36	"ticket": ticket,
37	"queue": "billing",
38	"summary": "Use the duplicate-charge runbook.",
39	}
40
41
42	@pure("ticket_prompt")
43	def ticket_prompt(hit: dict[str, str]) -> dict[str, str]:
44	return {"queue": hit["queue"], "context": hit["summary"]}
45
46
47	support_reply = Reasoner(
48	name="support_reply",
49	model="anthropic:claude-haiku-4-5-20251001",
50	system="Draft one concise support reply as JSON.",
51	reply=SupportReply,
52	)
53
54
55	@flow
56	def triage(ticket: str) -> dict[str, str]:
57	hit = lookup_ticket(ticket, retries=2, timeout_s=5)
58	prompt = ticket_prompt(hit)
59	answer = think(support_reply, prompt, timeout_s=10)
60	return hit \| answer
61
62
63	def fake_support_reply(value: dict[str, str]) -> SupportReply:
64	return {"reply": f"{value['queue']}: {value['context']}"}
65
66
67	deployment = deploy(triage, tools=[lookup_ticket], reasoners=[support_reply])
68	result = deployment.dry_run(
69	"Customer was charged twice.",
70	reasoners={"support_reply": fake_support_reply},
71	)
72
73	print(result.value)
74	```
75
76	`@flow` runs once at definition time with data handles. Registered tools, registered pures, `think(...)`, `cond(...)`, `switch(...)`, `each(...)`, and `reschedule(...)` append graph steps instead of doing runtime work; `\|` merges records and `h["key"]` plucks fields. `deploy(..., tools=..., reasoners=...)` freezes the tool and reasoner surface, and `dry_run(...)` executes locally with in-memory tools and deterministic fake reasoners. See the larger `@flow` examples in `examples/episode_summary_flow.py` and `examples/cluster_labeling_flow.py`.
77
78	---
79
80	## The CLI
81
82	`julep` is the developer CLI for a whole module of agents — "dbt for agents, terminal-native." Point it at a directory; it discovers every `@flow`/`Agent(...)`, treats each as a node in a cross-agent graph, and gives you one selection grammar across every verb:
83
84	```bash
85	julep ls # list agents (name · kind · tags)
86	julep show triage # one agent's kind, source location, tags, calls
87	julep graph # the cross-agent DAG as Graphviz DOT
88	julep run triage --input '"TICKET-42"' # execute locally, stream the trace tree
89	julep lint +triage # validate an agent and everything it depends on
90	julep test triage # run pytest for the selected agents
91	julep trace <run-id> # render a cached run's trace tree + Langfuse link
92	julep doctor # preflight: discovery, git, Langfuse, Temporal
93	julep deploy triage --env staging # freeze → publish → record in the deploy ledger
94	```
95
96	Selectors compose: `tag:support`, `state:modified` (Slim-CI), `+agent`/`agent+`/`@agent` graph traversal, `a,b` intersection, `--exclude`. Full reference: **[docs-site/content/doc

6 picks

Type

julep

How to install julep?

Related

julep

How to install julep?

Related